Marketers now anticipate festive needs, thanks to predictive commerce

What does it take for a marketer to map a consumer’s purchase journey in the age of quick commerce, where buying an iPhone — which usually involves months of planning and saving — can happen in a jiffy? This festive season, e-commerce has also observed unusual buying patterns, pushing marketers to re-examine the data at hand. For example, on Meesho, staples like jewellery and puja décor still led, but so did office supplies, sports equipment, wellness products, and books — many growing 60–100% year-on-year. Interestingly, gifting and daily needs have blurred into the same basket this season.

Now, what does this mean for platforms that track and predict consumer behaviour? This acceleration in purchase behaviour has created a new challenge for brands: identifying intent before it becomes obvious. Consider a user who spends weeks browsing high-end gadgets, then suddenly adds mid-range home décor to their cart. For a predictive model attuned to festive timing and gifting trends, this shift represpredictive modelents a transition from self-gifting to family gifting. Brands that catch this moment win relevance; those that miss it become noise.

Mapping an unpredictable consumer journey

India’s 2025 festive shopping season was bigger, faster, and broader than ever. Platforms saw staggering traffic. Meesho logged 206 crore visits, Flipkart’s early access doubled single-day orders, and Amazon drew over 38 crore visits in just two days, with most coming from beyond the top nine metros. Qcom boomed too, with orders up 85% year-on-year, ultra-fast deliveries, and booming new categories like home décor, wellness, and office supplies. Discounts remained aggressive, and 10-minute delivery alone is expected to generate $1.6 billion in sales, highlighting both the scale and speed of India’s evolving festive market.

And according to the Marketer’s Guide to India’s Festive Season by InMobi Advertising and Glance, festive shopping is becoming more and more digital every year. 64% of consumers plan to complete their festive shopping fully online, and 83% research digitally before buying. Discovery now happens across multiple touchpoints — social media (96%), in-app content (95%), OTT (92%), e-commerce sites (88%), and even lock screens (85%). Consistency matters too: 92% of shoppers are more likely to buy from brands they see repeatedly across mobile, CTV, and the web.

The shift toward digital has made predictive analytics indispensable. AI platform-related searches surged to 654 million in pre-festive 2025, a 2.6x jump from the previous year, according to GIPSI's GRWAi report. Nearly eight in ten consumers actively seek AI-powered recommendations, while 87% favour virtual try-ons for confidence in purchase decisions.

The consumer appetite for discovery extends to brands themselves. The early start to shopping behaviour has pushed advertisers to launch campaigns more than 30 days before Diwali.

Jacob Joseph, VP of Data Science at CleverTap, explains that identifying high-intent shoppers requires looking beyond demographics. "The art of prediction lies in reading intent before it's obvious. Festive buying isn't random — people drop subtle breadcrumbs weeks before the sale, from changes in browsing rhythm and wishlist behaviour to time spent exploring certain categories," he notes. "Predictive models work best when they focus less on demographics and more on moment-based signals — recency, engagement depth, and repeat interactions — cues that reveal who's emotionally primed to buy long before price comparisons begin."

The data signals fed into these models vary widely. Vishal Shrivastava, Head of Business Strategy at AnyMind Group India, emphasises the importance of early signals in markets like India. "With culturally rich countries like India, it is very important to ensure that we, as enablers, have the readiness to rightfully use predictive analytics by feeding as much past data as possible," he explains.

Key festivals like Diwali generate anticipation from the consumer standpoint weeks or even months in advance. Shrivastava points to users who actively engage with rewarded videos or play mini-games within mobile ad placements as showing a high-attention, receptive mindset. “When predictive models combine this signal with historical AOV data signal from our data partnership platforms, it flags a High-Intent Shopper weeks before they've performed a single promotional search.” It allows the platform to prioritise an audience segment that is demonstrably ready to engage, driving efficiency and minimising wasted ad spend.

For festive campaigns specifically, the models blend historical festive data with real-time inputs. Joseph describes the process as dynamic learning with systems that absorb past trends while listening to the present. This includes browsing activity, engagement spikes, gifting-related searches, and regional search patterns. 

Raahul Seshadri, Director of AI and Tech at WebEngage, outlines, "A mix of historical purchase data, category preferences, and seasonal behaviour patterns, alongside real-time signals like browsing history, location, engagement with past campaigns, and even social media interactions, are blended together.” 

The key lies in knowing which signals truly indicate readiness to buy, while past festive trends reveal what people are likely to purchase, and real-time behaviour helps understand when they are ready to act.

When patterns flag purchase intent 

Predictive models during festivals go beyond pattern recognition. Signals like gift searches, browsing frequency, cart abandonment, and regional spending trends are key but even subtle cues, such as engagement with Diwali promotions or messaging platform interactions, help refine targeting.

Performance trends from last year’s festive season highlight how different formats respond to predictive optimisation. Full-screen playable ads delivered 1.5x higher click-through rates, and CTV banners drove 1.3x more engagement than standard formats. Predictive models are finding the right audience and identifying which creative formats work best at each stage of the purchase journey.

Shrivastava outlines how AnyMind Group's mobile marketing platform, POKKT, captures what he calls Interaction Intent. The platform uses non-disruptive, rich-media ads placed within mobile apps to gather deep behavioural data that traditional platforms miss. 

"We don't just rely on transactional history; we use POKKT to capture Interaction Intent, the voluntary engagement with non-disruptive, rich-media ads placed within mobile apps," he explains. 

For example, a predictive model might determine that for the pre-Diwali home preparation phase, a gamified home décor quiz delivered via a POKKT interstitial ad will perform 40% better than a static banner, according to him.

The crucial data point is the Interaction Intent Score: whether users tap to view a 3D model, spin the wheel for a discount, or engage with other interactive elements. This active participation provides a direct, measurable signal of conversion intent that passive ad formats cannot capture.

On-device advertising is proving the power of predictive targeting. The VEVE Festive Report 2025 highlights splash screens, notifications, app stores, and lock screens as tools to reach consumers at micro-moments. AJIO achieved over 25x return on ad spend using browser and lock screen formats, while BFSI brands like ACKO saw an average 18% conversion through strategic app install campaigns on vivo and Xiaomi native app stores.

Seshadri cites Coca-Cola’s collaboration with Paytm Mall during Diwali in 2025, where the brand integrated cashback rewards into its campaign, leveraging predictive marketing to identify high-intent shoppers and personalise offers. The approach resonated emotionally with consumers and led to a significant increase in sales and brand engagement during the festive period.

In another case, AnyMind Group worked with a brand targeting users showing early festive intent, those with high mobile usage and high-value shopping history. Instead of deploying a simple video ad, the platform used a POKKT Rich Media interstitial featuring a Diwali-themed interactive challenge. The ad included a 'Spin-to-Win' discount and a dynamic 3D view of the festive-edition model. 

Shrivastava explains that by making the ad an event, it leveraged data captured by POKKT, specifically Interaction Rate and Dwell Time, as an immediate predictor of high-quality leads. 

"Users who actively engaged for more than 15 seconds were instantly prioritised and funnelled directly to a personalised landing page. This approach resulted in a significant double-digit reduction in CPL compared to their standard social video campaigns,” he notes. 

Festive marketing, but in the right context

The challenge of balancing hyper-personalisation with the scale of Diwali audiences becomes acute in a market as diverse as India. MiQ’s Festive Shopper Insights 2025 shows clear regional differences: North and Central India favour automobiles, gadgets, and appliances, while South and East India show strong rural demand for two-wheelers, tractors, and jewellery despite rising gold prices. Urban areas saw softer early demand across categories, though gold remained a preferred investment-led purchase.

Joseph notes that brands need to focus on individualisation rather than personalisation at scale when using predictive analytics. 

"Personalisation at scale isn't about crafting millions of messages, but understanding millions of contexts," he explains. "India's festive landscape varies widely in sentiment, timing, and buying power across regions. The way forward is individualisation — campaigns that respond to intent clusters rather than static personas. The emotional core of joy, generosity, and togetherness can remain consistent, but the expression adapts to each context, letting brands speak the same language as their audience."

Shrivastava notes that hyper-personalisation in India is comparatively challenging compared to Western markets due to the many languages, cultures, and traditions. The strategy moves beyond simple demographics (age/gender) into cultural intent and vernacular affinity. “We map regional purchase traditions (e.g., gold buying on Dhanteras vs. specific regional sweet hampers) against behavioural data," he explains. The AI system uses this map to instantly localise creative and offers, often leveraging insights from local creators whose voices resonate within specific communities. This is to ensure communication feels designed for a local celebration, achieving hyper-relevance at a national scale, giving a huge audience the feeling of a personalised, familiar, and highly local shopping experience.

Seshadri stresses the need for adaptable segments that respect regional differences while personalising messages based on individual behaviour, making campaigns relevant without overwhelming shoppers. With real-time data and AI insights, brands can anticipate trends, customise offers, and optimise campaigns dynamically.

The shift extends beyond current capabilities. Joseph sees predictive commerce moving festive marketing from reactive engagement to anticipatory marketing. 

"Looking ahead, predictive commerce will shift festive marketing from reactive engagement to anticipatory marketing," he notes. "Innovation won't just be predicting what people will buy, but understanding why; the emotions, convenience, and sense of belonging driving their choices. That combination of foresight and empathy is where the next wave of festive marketing breakthroughs will emerge."

Shrivastava envisions predictive commerce fundamentally shifting toward what he calls Agentic Commerce over the next two to three years. “The AI system will stop being an advisor and start being an executor, managing autonomous operations across the commerce value chain," he explains. This will result in two major shifts: Supply Chain as a Marketing Lever, where prediction directly informs inventory movement for quick-commerce efficiency, and the rise of Service-as-a-Product, where AI anticipates customer anxiety and automatically triggers high-value, bespoke service offers like home-prep consultations bundled with purchases.

Seshadri expects campaigns to become almost intuitive in the coming years. "In the next 2-3 years, we'll see campaigns that feel almost intuitive, delivering the right message, to the right shopper, at the exact right moment, making the shopping experience completely smooth and enjoyable for both brands and consumers."

The shift from demographic targeting to moment-based intent detection, from reactive campaigns to anticipatory engagement, marks a fundamental change in the relationship between brands and shoppers. Festive marketing is no longer about catching attention, it’s about reading intent early enough to meet consumers where their desires begin.